*For correspondence: sheltzer@

cshl.edu

†These authors contributed

equally to this work

Competing interests: The

authors declare that no

competing interests exist.

Funding: See page 14

Received: 12 December 2016

Accepted: 20 February 2017

Published: 24 March 2017

Reviewing editor: Jeffrey

Settleman, Calico Life Sciences,

United States

Copyright Lin et al. This article

is distributed under the terms of

the Creative Commons

Attribution License, which

permits unrestricted use and

redistribution provided that the

original author and source are

credited.

CRISPR/Cas9 mutagenesis invalidates aputative cancer dependency targeted inon-going clinical trialsAnn Lin1,2†, Christopher J Giuliano1,2†, Nicole M Sayles1, Jason M Sheltzer1*

1Cold Spring Harbor Laboratory, Cold Spring Harbor, United States; 2Stony BrookUniversity, Stony Brook, United States

Abstract The Maternal Embryonic Leucine Zipper Kinase (MELK) has been reported to be a

genetic dependency in several cancer types. MELK RNAi and small-molecule inhibitors of MELK

block the proliferation of various cancer cell lines, and MELK knockdown has been described as

particularly effective against the highly-aggressive basal/triple-negative subtype of breast cancer.

Based on these preclinical results, the MELK inhibitor OTS167 is currently being tested as a novel

chemotherapy agent in several clinical trials. Here, we report that mutagenizing MELK with

CRISPR/Cas9 has no effect on the fitness of basal breast cancer cell lines or cell lines from six other

cancer types. Cells that harbor null mutations in MELK exhibit wild-type doubling times,

cytokinesis, and anchorage-independent growth. Furthermore, MELK-knockout lines remain

sensitive to OTS167, suggesting that this drug blocks cell division through an off-target

mechanism. In total, our results undermine the rationale for a series of current clinical trials and

provide an experimental approach for the use of CRISPR/Cas9 in preclinical target validation that

can be broadly applied.

DOI: 10.7554/eLife.24179.001

IntroductionTumors of the breast can be divided into five distinct subtypes based on characteristic gene expres-

sion patterns. These breast cancer subtypes are referred to as Luminal A, Luminal B, Her2-enriched,

normal-like, and basal (Sørlie et al., 2001). Basal breast cancers (BBCs) comprise ~15% of all diag-

nosed breast cancers and express genes typically found in the basal/myoepithelial layer of the mam-

mary gland (Badve et al., 2011; Rakha et al., 2008). BBCs are most frequently diagnosed in

younger patients and present with advanced histologic grade, central necrosis, and high mitotic

activity. Additionally, ~70% of basal breast cancers fail to express the estrogen receptor (ER), the

progesterone receptor (PR), or the human epidermal growth factor receptor 2 (HER2) (Badve et al.,

2011). Tumors that lack expression of ER, PR, or HER2 are referred to as ‘triple-negative’ breast can-

cers, and are irresponsive to hormonal or anti-HER2 therapies that have proven effective against

receptor-positive cancers. Due to their resistance to targeted therapies as well as their rapid rate of

cell division, basal breast cancers currently have the worst prognosis of any breast cancer subtype.

Thus, there is an urgent need to develop new therapies that are effective against triple-negative or

basal-type tumors.

In recent years, significant progress has been made in the treatment of certain malignancies by

targeting cancer cell ‘addictions’, or genetic dependencies that encode proteins required for the

growth of specific cancer types (Luo et al., 2009). Drugs that block the function of a cancer depen-

dency – like the antibody Herceptin in Her2+ breast cancer – can trigger apoptosis and durable

tumor regression (Weinstein, 2002). Cancer cell addictions are often investigated through the use

of different transgenic technologies to disrupt the expression of a specified gene. Two of the most

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 1 of 17

SHORT REPORT

popular methodologies are RNA interference, which destabilizes a targeted transcript, and CRISPR

mutagenesis, which utilizes the nuclease Cas9 to induce frameshift mutations at a targeted locus.

While CRISPR-mediated genetic engineering has been widely adopted since its discovery in 2013,

RNA interference remains popular due to its ability to deplete multiple isoforms of a protein, its

reversibility, and its relative insensitivity to gene copy number (Boettcher and McManus, 2015).

Moreover, the partial loss-of-function phenotype generated by RNAi may more accurately recapitu-

late the effects of drug treatment than the complete loss-of-function phenotype generated by a

Cas9-induced frameshift mutation. Nonetheless, RNAi constructs exhibit limited specificity, and off-

target knockdowns are an inherent and widespread problem in RNAi experiments (Jackson et al.,

2003, 2006; Singh et al., 2011).

The Maternal Embryonic Leucine Zipper Kinase (MELK) has received substantial attention as a

potential cancer cell addiction and promising target for drug development. MELK was first identified

as an AMPK family member expressed in the mouse pre-implantation embryo (Heyer et al., 1997),

and has since been implicated in several cellular processes, including apoptosis (Jung et al., 2008),

splicing (Vulsteke et al., 2004), and neurogenesis (Nakano et al., 2005). MELK is also over-

expressed in most types of solid tumors, including breast, colon, liver, lung, melanoma, and ovarian

cancer (Gray et al., 2005). Furthermore, many publications have reported that knocking down

MELK using RNAi inhibited the proliferation of cell lines derived from these cancer types

(Gray et al., 2005; Lin et al., 2007; Kuner et al., 2013; Du et al., 2014; Kig et al., 2013;

Speers et al., 2016; Alachkar et al., 2014; Marie et al., 2008; Nakano et al., 2008;

Hebbard et al., 2010; Wang et al., 2014; Choi et al., 2011; Xia et al., 2016; Gu et al., 2013). In

particular, MELK has been identified as a key driver of basal-type breast cancer, suggesting a novel

therapeutic approach to treat this disease (Wang et al., 2014). In response to the widespread

reports that MELK is a cancer dependency, several companies have developed small molecule inhibi-

tors of MELK that block the activity of the kinase in vitro and that inhibit cancer cell proliferation at

micromolar or nanomolar concentrations (Beke et al., 2015; Toure et al., 2016; Johnson et al.,

2015a, 2015b; Chung et al., 2012). Additionally, four clinical trials have been launched to test the

MELK inhibitor OTS167 in human cancers (NCT01910545, NCT02768519, NCT02795520, and

NCT02926690).

As part of a project in our lab to characterize genes whose expression is associated with patient

prognosis in cancer (Sheltzer, 2013), we identified MELK as highly-expressed in deadly tumors from

multiple cancer types (data not shown). We set out to use CRISPR/Cas9 to characterize the effects of

eLife digest Like a person who is dependent on coffee to be productive, cancer cells are

dependent on the products of certain genes in order to dominate their environment and grow.

Cancer cells will stop growing and die when the activity of these gene products is blocked. These

genes are known as cancer dependencies or “addictions”. As a result, researchers are constantly

looking for cancer dependencies and developing drugs to block their activity.

It was previously believed that a gene called MELK was an addiction in certain types of breast

cancer. In fact, pharmaceutical companies had developed a drug to block the activity of MELK, and

this drug is currently being tested in human patients. However, Lin, Giuliano et al. have now taken a

second look at the role of MELK in breast cancer, and have come to a different conclusion.

Using a gene editing technology called CRISPR/Cas9, Lin, Giuliano et al. removed MELK activity

from several cancer cell lines. This did not stop cancer cells from multiplying, suggesting that MELK

is not actually a cancer addiction.

Additionally, when breast cancer cells that do not produce MELK were exposed to the drug that

is supposed to block MELK activity, the drug still stopped cell growth. Since the drug works when

MELK is not present in the cell, the drug must be binding to other proteins. This suggests that MELK

is not the actual target of the drug.

Lin, Giuliano et al. suggest that, in the future, CRISPR/Cas9 technology could be used to better

identify cancer dependencies and drug targets before cancer drugs are given to human patients.

DOI: 10.7554/eLife.24179.002

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 2 of 17

Short report Cancer Biology Genes and Chromosomes

MELK loss on tumorigenesis. Unexpectedly, we found that mutating MELK failed to affect the

growth of every cancer cell line that we tested. Furthermore, the MELK inhibitor OTS167 remained

effective against cells with null mutations in MELK, suggesting that its in vivo activity results from an

off-target effect. We propose that CRISPR represents an essential modality to confirm putative can-

cer dependencies and drug specificity before preclinical findings are advanced to human trials.

Results

Mutagenizing MELK using CRISPR/Cas9Over a dozen previous publications have reported that MELK is a cancer dependency, as blocking

MELK with RNAi or small molecules inhibited the proliferation of cell lines derived from multiple

tumor types (Gray et al., 2005; Lin et al., 2007; Kuner et al., 2013; Du et al., 2014; Kig et al.,

2013; Speers et al., 2016; Alachkar et al., 2014; Marie et al., 2008; Nakano et al., 2008;

Hebbard et al., 2010; Wang et al., 2014; Choi et al., 2011; Xia et al., 2016; Gu et al., 2013).

However, several discrepancies exist in the literature on MELK. For instance, various publications dis-

agree over the cell cycle stage affected by MELK inhibition (Du et al., 2014; Kig et al., 2013;

Alachkar et al., 2014; Wang et al., 2014; Beke et al., 2015), while other publications disagree over

whether receptor-positive breast cancer cell lines are sensitive (Lin et al., 2007; Beke et al., 2015;

Chung et al., 2012) or resistant (Wang et al., 2014) to MELK inhibition. To unambiguously deter-

mine the effects of MELK loss in cancer cell lines, we applied CRISPR/Cas9 to generate frameshift

mutations in the MELK coding sequence. We designed seven guide RNAs (gRNAs) against MELK,

five of which target the N-terminal kinase domain and two of which target the C-terminal kinase-

associated domain (Figure 1A and Supplementary file 1). Then, we cloned each guide RNA into a

GFP-expressing vector and transduced the guides into three Cas9-expressing cell lines: the triple-

negative breast cancer cell lines Cal51 and MDA-MB-231, reported to be addicted to MELK expres-

sion (Wang et al., 2014), and the melanoma cell line A375, from a cancer type that over-expresses

MELK (Gray et al., 2005; Ryu et al., 2007). As negative controls in these assays, we also cloned and

transduced three gRNA’s that target the non-essential and non-coding Rosa26 locus

(Supplementary file 1).

To assess the efficacy of our CRISPR system, we purified GFP+ populations from cell lines harbor-

ing each individual guide RNA, and then we analyzed the targeted loci by Sanger sequencing. TIDE

analysis, which decomposes raw sequencing traces into linear combinations of indel mutations

(Brinkman et al., 2014), revealed high cutting efficiency at most targeted loci (Figure 1B and Fig-

ure 1—figure supplements 1–3). Across the 21 samples, the median level of indel formation was

80%. These values likely underestimate the true mutation frequency, as TIDE analysis is not able to

detect missense mutations and large indels will not be efficiently amplified by PCR. Western blot

analysis of MELK protein levels in A375, Cal51, and MDA-MB-231 sorted populations further con-

firmed that our CRISPR system effectively ablated MELK expression (Figure 1C and Figure 1—fig-

ure supplement 4).

We next set out to determine whether MELK was required for cancer cell fitness. To test this, we

measured cell proliferation over 15 days in culture in the 30 independent lines of A375, Cal51, and

MDA-MB-231 that we had generated (Figure 1D–F). Surprisingly, we failed to detect any difference

in proliferative capacity between the cell lines with wild-type or mutant MELK. For instance, in the

A375 cell line, we calculated a mean doubling time of 16.9 hr among cells transduced with Rosa26

guide RNAs, while the cell lines transduced with MELK gRNAs exhibited a mean doubling time of

16.8 hr. MELK gRNA-transduced cell lines also exhibited wild-type levels of growth under anchor-

age-independent conditions (Figure 1G–I). These results call into question the notion that MELK is a

genetic dependency either across cancer types or in triple-negative breast cancers.

MELK is not a common cancer cell dependencyIn order to assess whether a wider range of cancer cell lines were dependent on MELK for viability,

we performed individual GFP dropout experiments in 13 Cas9-expressing cancer cell lines

(Shi et al., 2015). In these assays, cancer cells are transduced with GFP-expressing guide RNA vec-

tors at low MOI to create mixed populations of GFP+ and GFP- cells. A guide RNA that induces

mutations in a gene required for cancer cell fitness will drop out from the population, resulting in a

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 3 of 17

Short report Cancer Biology Genes and Chromosomes

Rosa

26 g

RNA

MELK

gRNA

0

5

10

15

20

Do

ub

lin

g t

ime (

ho

urs

)

A375: Doubling Time

0 1 2 3 4 50

5

10

15

20

25

Passage

Po

pu

lati

on

Do

ub

lin

gs

A375: ProliferationRosa26 g1

Rosa26 g2

Rosa26 g3

MELK g1

MELK g2

MELK g3

MELK g4

MELK g5

MELK g6

MELK g7

A

Rosa

26 g

RNA

MELK

gRNA

0

5

10

15

20

25

Do

ub

lin

g t

ime (

ho

urs

)

Cal51: Doubling Time

0 1 2 3 4 50

5

10

15

20

Passage

Po

pu

lati

on

Do

ub

lin

gs

Cal51: ProliferationRosa26 g1

Rosa26 g2

Rosa26 g3

MELK g1

MELK g2

MELK g3

MELK g4

MELK g5

MELK g6

MELK g7

Rosa

26 g

RNA

MELK

gRNA

0

10

20

30

40

Do

ub

lin

g t

ime (

ho

urs

)

MDA-MB-231: Doubling Time

0 1 2 3 4 50

5

10

15

Passage

Po

pu

lati

on

Do

ub

lin

gs

MDA-MB-231: ProliferationRosa26 g1

Rosa26 g2

Rosa26 g3

MELK g1

MELK g2

MELK g3

MELK g4

MELK g5

MELK g6

MELK g7

Rosa26 gRNA MELK gRNA % In-del

MELK g1 78.2%

MELK g2 85.3%

MELK g3 84.2%

Ro

sa

26

g1

Ro

sa

26

g2

ME

LK

g1

ME

LK

g2

ME

LK

g3

ME

LK

g4

ME

LK

g5

ME

LK

g6

ME

LK

g7

MELK

α-TUB

Rosa26 g1 MELK g3 MELK g6

So

ft A

ga

r

Cal51: anchorage-independent growth

Rosa

26 g

1

Rosa

26 g

2

Rosa

26 g

3

MELK

g1

MELK

g2

MELK

g3

MELK

g4

MELK

g5

MELK

g6

MELK

g7

0

2

4

6

8

Co

lon

ies p

er

field

B

C

D

E

F

G

H

IA375: anchorage-independent growth

Rosa

26 g

1

Rosa

26 g

2

Rosa

26 g

3

MELK

g1

MELK

g2

MELK

g3

MELK

g4

MELK

g5

MELK

g6

MELK

g7

0

2

4

6

8

10

Co

lon

ies p

er

field

Figure 1. Mutation of the MELK kinase domain does not affect cancer cell proliferation or anchorage-independent growth. (A) Domain structure of

MELK and locations of the sequences targeted by 7 MELK gRNAs. (B) Genomic DNA was purified from the indicated population of MDA-MB-231 cells

and the targeted loci were amplified by PCR. Percent indel formation was estimated using TIDE analysis. The highlighted region indicates 20

nucleotides or 15 nucleotides of the sequence recognized by the guide RNA. Other sequence traces are presented in Figure 1—figure supplements

1–3. (C) Western blot analysis of GFP+ MDA-MB-231 cells using the Abcam ab108529 MELK antibody. Alpha-tubulin levels were analyzed as a loading

control. (D–F) Proliferation and doubling time analysis of A375, Cal51, and MDA-MB-231 cell lines transduced with 3 Rosa26 gRNAs or with 7 MELK

gRNAs. (G) Images of colonies from the indicated Cal51 strains grown in soft agar. (H–I) Quantification of anchorage-independent growth in Cal51 or

Figure 1 continued on next page

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 4 of 17

Short report Cancer Biology Genes and Chromosomes

decreasing ratio of GFP+ to GFP- cells over time. Additionally, we considered it possible that cells

had adapted to MELK loss during the time required to sort and expand pure GFP+ cell populations

to perform the experiments described in Figure 1. For the following dropout assays, we monitored

GFP levels directly following introduction of the gRNA virus, without selecting or expanding cell

populations. Importantly, this strategy allows us to detect whether the mutation of MELK results in a

transient or immediate loss of cell fitness.

As negative controls in this experiment, we utilized three gRNA’s targeting Rosa26, and as posi-

tive controls we designed six gRNA’s targeting the essential replication genes RPA3 and PCNA

(Supplementary file 1). We first transduced these gRNA’s individually into seven triple-negative

breast cancer cell lines (Cal51, HCC1143, HCC1937, HCC70, MDA-MB-231, MDA-MB-453, and

MDA-MB-468). Over the course of five passages in culture, gRNA’s targeting Rosa26 typically

depleted 1.2 to 2-fold (Figure 2A). This low level of depletion may result from off-target mutagene-

sis or from cell cycle arrest caused by repeated DNA breaks (Aguirre et al., 2016). Over the same

period of time, gRNA’s targeting RPA3 and PCNA depleted 5-fold to 100-fold. These positive con-

trol guides exhibited varying degrees of dropout (e.g., compare RPA3 g1 and RPA3 g2), which may

result from variability in cutting efficiency or from functionally-important differences in the protein

domains targeted by these guides. However, in every cell line tested, every single gRNA targeting

RPA3 or PCNA dropped out to a greater degree than every single Rosa26 guide. In contrast to

RPA3 and PCNA, the seven guides that targeted MELK typically depleted less than 2-fold. Across

seven different gRNAs tested in seven different cell lines, we never observed a MELK guide deplete

more than 2.5-fold. In six of the seven cell lines, a Rosa26 gRNA exhibited a higher level of depletion

than every single MELK gRNA (Figure 2B). We conclude that these seven triple-negative breast can-

cer cell lines are not dependent on MELK for cell fitness.

To extend these observations, we repeated the GFP dropout experiments in six Cas9-expressing

cell lines (A375, Cama1, HCT116, NCI-H1299, T24, and U118-MG) from other cancer types previ-

ously suggested to require MELK expression. Consistent with our observations in the triple-negative

breast cancer cells, guides targeting Rosa26 exhibited minimal dropout over five passages, while

guides targeting RPA3 or PCNA were depleted up to 90-fold (Figure 2—figure supplement 1).

However, 0 of the 7 MELK guides exhibited significant dropout in any of the cell lines tested (maxi-

mum dropout: 1.8-fold). Again, guides targeting Rosa26 exhibited an equivalent or occasionally

greater degree of depletion than guides targeting MELK (Figure 2—figure supplement 1B). In

total, this data suggests that MELK is not a common cancer dependency.

Unbiased RNAi and CRISPR screens fail to identify MELK as a cancerdependencySeveral laboratories have conducted genome-wide or kinase-focused screens to identify novel can-

cer addictions. If these unbiased screens indicated that cancer cell lines required MELK expression

to proliferate, then that would bolster the contention that MELK could be a therapeutic target in

cancer. We therefore examined data from four recent screens: a kinome-wide siRNA screen in 117

cancer cell lines (Campbell et al., 2016), a genome-wide CRISPR screen in 6 cell lines (Hart et al.,

2015), a genome-wide shRNA screen in 72 cancer cell lines (Marcotte et al., 2012; Hart et al.,

Figure 1 continued

A375 cells transduced with the indicated gRNA. For each assay, colonies were counted in at least 15 fields under a 10x objective. Boxes represent the

25th, 50th, and 75th percentiles of colonies per field, while the whiskers represent the 10th and 90th percentiles.

DOI: 10.7554/eLife.24179.003

The following figure supplements are available for figure 1:

Figure supplement 1. Mutation of MELK using seven different guide RNAs in the A375 melanoma cell line.

DOI: 10.7554/eLife.24179.004

Figure supplement 2. Mutation of MELK using seven different guide RNAs in the Cal51 triple-negative breast cancer cell line.

DOI: 10.7554/eLife.24179.005

Figure supplement 3. Mutation of MELK using seven different guide RNAs in the MDA-MB-231 triple-negative breast cancer cell line.

DOI: 10.7554/eLife.24179.006

Figure supplement 4. Western blot analysis of MELK-disrupted cell populations.

DOI: 10.7554/eLife.24179.007

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 5 of 17

Short report Cancer Biology Genes and Chromosomes

Rosa

26 g

1

Rosa

26 g

2

Rosa

26 g

3

RPA

3 g1

RPA

3 g2

RPA

3 g3

PCNA g

1

PCNA g

2

PCNA g

3

MELK

g1

MELK

g2

MELK

g3

MELK

g4

MELK

g5

MELK

g6

MELK

g7

0

10

20

30

40

50

Fo

ld d

ep

leti

on

HCC1143

Passage 1

Passage 2

Passage 3

Passage 4

Passage 5

Rosa

26 g

1

Rosa

26 g

2

Rosa

26 g

3

RPA

3 g1

RPA

3 g2

RPA

3 g3

PCNA g

1

PCNA g

2

PCNA g

3

MELK

g1

MELK

g2

MELK

g3

MELK

g4

MELK

g5

MELK

g6

MELK

g7

0

10

20

30

Fo

ld d

ep

leti

on

MDA-MB-453

Passage 1

Passage 2

Passage 3

Passage 4

Passage 5

Rosa

26 g

1

Rosa

26 g

2

Rosa

26 g

3

RPA

3 g1

RPA

3 g2

RPA

3 g3

PCNA g

1

PCNA g

2

PCNA g

3

MELK

g1

MELK

g2

MELK

g3

MELK

g4

MELK

g5

MELK

g6

MELK

g7

0

20

40

60

80

100

Fo

ld d

ep

leti

on

MDA-MB-231

Passage 1

Passage 2

Passage 3

Passage 4

Passage 5

Rosa

26 g

1

Rosa

26 g

2

Rosa

26 g

3

RPA

3 g1

RPA

3 g2

RPA

3 g3

PCNA g

1

PCNA g

2

PCNA g

3

MELK

g1

MELK

g2

MELK

g3

MELK

g4

MELK

g5

MELK

g6

MELK

g7

0

10

20

30

Fo

ld d

ep

leti

on

MDA-MB-468

Passage 1

Passage 2

Passage 3

Passage 4

Passage 5

Rosa

26 g

1

Rosa

26 g

2

Rosa

26 g

3

RPA

3 g1

RPA

3 g2

RPA

3 g3

PCNA g

1

PCNA g

2

PCNA g

3

MELK

g1

MELK

g2

MELK

g3

MELK

g4

MELK

g5

MELK

g6

MELK

g7

0

5

10

15

20

25F

old

dep

leti

on

CAL51

Passage 1

Passage 2

Passage 3

Passage 4

Passage 5

Rosa

26 g

1

Rosa

26 g

2

Rosa

26 g

3

RPA

3 g1

RPA

3 g2

RPA

3 g3

PCNA g

1

PCNA g

2

PCNA g

3

MELK

g1

MELK

g2

MELK

g3

MELK

g4

MELK

g5

MELK

g6

MELK

g7

0

5

10

15

Fo

ld d

ep

leti

on

HCC1937

Passage 1

Passage 2

Passage 3

Passage 4

Passage 5

Rosa

26 g

1

Rosa

26 g

2

Rosa

26 g

3

RPA

3 g1

RPA

3 g2

RPA

3 g3

PCNA g

1

PCNA g

2

PCNA g

3

MELK

g1

MELK

g2

MELK

g3

MELK

g4

MELK

g5

MELK

g6

MELK

g7

0

5

10

15

20

25

Fo

ld d

ep

leti

on

HCC70

Passage 1

Passage 2

Passage 3

Passage 4

Passage 5

A

B

Figure 2. Guide RNAs targeting MELK fail to drop out in triple-negative breast cancer cell line competition experiments. (A) The fold change in the

percentage of GFP+ cells, relative to the percentage of GFP+ cells at passage 1, is displayed for seven triple-negative breast cancer cell lines. (B) A

table summarizing the results presented in (A) is displayed.

DOI: 10.7554/eLife.24179.008

Figure 2 continued on next page

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 6 of 17

Short report Cancer Biology Genes and Chromosomes

2014), and a genome-wide CRISPR screen in seven cell lines (Tzelepis et al., 2016). Large-scale

unbiased screens are prone to experimental artifacts, and variations in protocol, technology, or the

method of analysis can cause different screens to yield different results (Mohr et al., 2014). None-

theless, each of these screens identified multiple mitotic kinases as essential in various cancer cell

lines, including Aurora B, BubR1, CDK1, and Plk1 (Figure 2—figure supplement 2). In contrast,

MELK was not identified as essential in a single experiment. These negative results include 13 triple-

negative breast cancer cell lines tested by Campbell et al. and 16 triple-negative breast cancer cell

lines tested by the Moffatt lab. Several other published pan-cancer or breast cancer-focused screens

have failed to identify MELK as either a general cancer dependency or a triple-negative breast can-

cer dependency (Silva et al., 2008; Marcotte et al., 2016; Cowley et al., 2014). Thus, while we do

not consider results from large-scale screens to be dispositive, we believe that these findings, cou-

pled with our own experimental evidence, suggest that MELK expression is not required for cancer

cell proliferation.

OTS167 inhibits the growth of receptor-positive breast cancer cell linesand cells that harbor mutant MELKIf MELK is not a cancer cell dependency, then drugs that inhibit MELK must either be ineffective at

stopping cancer cell division or they must also act on other cellular targets. We therefore assessed

the efficacy of the MELK inhibitor OTS167 (alternately called OTSSP167), a therapeutic agent being

tested in several clinical trials. We treated a variety of cancer cell lines with 7-point serial dilutions of

OTS167, and we observed that OTS167 did in fact impede cell proliferation at nanomolar concentra-

tions (mean GI50 = 16 nM; see below). As OTS167 was able to inhibit growth despite the non-essen-

tially of MELK, we considered the possibility that OTS167 acted through an off-target effect. To test

this, we set out to determine whether MELK expression was actually required for OTS167 sensitivity.

We calculated the GI50 value of OTS167 in A375, Cal51, and MDA-MB-231 cells that harbored

gRNA’s targeting either Rosa26 or MELK, and we found that cell populations with wild-type or

mutant MELK displayed equivalent sensitivity to the drug (Figure 3). For instance, in Cal51 cells

transduced with Rosa26 gRNAs, the GI50 values ranged from 9 nM to 12 nM (mean: 10 nM), while in

Cal51 cells transduced with MELK gRNAs, the GI50 values ranged from 8 nM to 14 nM (mean: 11

nM). As OTS167 exhibits nanomolar potency against cancer cell lines but is unaffected by mutations

in MELK, this suggests that OTS167 blocks proliferation by inhibiting another target or targets.

To further explore this observation, we tested the efficacy of OTS167 against a panel of triple-

negative or receptor-positive breast cancer cell lines. MELK is significantly up-regulated in triple-

negative tumors relative to receptor-positive tumors (Wang et al., 2014), and one clinical trial

(NCT02926690) includes a dosage-escalation study of OTS167 in patients with triple-negative can-

cers. However, we observed no significant difference between the GI50 values of OTS167 according

to receptor status (Figure 3—figure supplement 1). In triple-negative breast cancer cell lines,

OTS167 inhibited growth by 50% at concentrations ranging from 10 nM to 42 nM (mean: 19 nM),

while in receptor-positive breast cancer cells GI50 values ranged from 9 nM to 21 nM (mean: 14

nM). These results demonstrate that OTS167 is not specifically effective against triple-negative

breast cancer cell lines, but instead remains remarkably potent against breast cancer cell lines that

express hormone receptors.

Lastly, to confirm that OTS167 treatment fails to phenocopy MELK mutations, we examined their

effects on cell cycle progression. We found that treatment with OTS167 blocked cytokinesis in a

dose-dependent manner, resulting in populations harboring 13% to 60% multinucleate cells. In con-

trast, cells transduced with either Rosa26 or MELK gRNAs progressed through the cell cycle without

gross mitotic defects, and exhibited no significant difference in the frequency of multinucleate cells

Figure 2 continued

The following figure supplements are available for figure 2:

Figure supplement 1. Guide RNAs targeting MELK fail to drop out in several cancer cell lines.

DOI: 10.7554/eLife.24179.009

Figure supplement 2. Unbiased screens do not identify MELK as a cancer dependency.

DOI: 10.7554/eLife.24179.010

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 7 of 17

Short report Cancer Biology Genes and Chromosomes

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

A375 MELK g5

GI50 = 14nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

A375 MELK g4

GI50 = 13nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

A375 Rosa26 g2

GI50 = 20nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

A375 MELK g7

GI50 = 13nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

A375 MELK g6

GI50 = 13nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

A375 Rosa26 g3

GI50 = 18nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

A375 MELK g2

GI50 = 20nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

A375 MELK g1

GI50 = 18nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

A375 Rosa26 g1

GI50 = 14nM

Rosa

26 g

RNA

MELK

gRNA

0

5

10

15

20

25

GI5

0 v

alu

es (

nM

)

A375: OTSSP167 sensivity

A B

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

Cal51 Rosa26 g1

GI50 = 10nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

Cal51 Rosa26 g2

GI50 = 12nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

Cal51 MELK g1

GI50 = 12nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

Cal51 MELK g2

GI50 = 14nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

Cal51 MELK g4

GI50 = 11nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

Cal51 MELK g5

GI50 = 12nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

Cal51 Rosa26 g3

GI50 = 9nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

Cal51 MELK g6

GI50 = 8nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

Cal51 MELK g7

GI50 = 9nMRosa

26 g

RNA

MELK

gRNA

0

5

10

15

20

GI5

0 v

alu

es (

nM

)

Cal51: OTSSP167 sensitivity

C D

Rosa

26 g

RNA

MELK

gRNA

0

5

10

15

20

GI5

0 v

alu

es (

nM

)

MDA-MB-231: OTSSP167 sensitivity

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 Rosa26 g1

GI50 = 18nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 Rosa26 g2

GI50 = 15nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 Rosa26 g3

GI50 = 16nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 MELK g1

GI50 = 13nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 MELK g2

GI50 = 16nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 MELK g3

GI50 = 18nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 MELK g4

GI50 = 13nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 MELK g6

GI50 = 18nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 MELK g7

GI50 = 15nM

E F

Figure 3. Mutating MELK does not affect OTS167 sensitivity. (A) Summary of GI50 values from OTS167 treatment of A375 cells harboring guide RNAs

targeting Rosa26 or MELK. (B) 7 point dose-response curves of OTS167 in the indicated A375 cell lines. (C) Summary of GI50 values from OTS167

treatment of Cal51 cells harboring guide RNAs targeting Rosa26 or MELK. (D) 7 point dose-response curves of OTS167 in the indicated Cal51 cell lines.

Figure 3 continued on next page

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 8 of 17

Short report Cancer Biology Genes and Chromosomes

(Figure 3—figure supplement 2). These results demonstrate that OTS167 induces a cell cycle failure

phenotype that is not recapitulated by mutagenizing MELK. We note that this observation is consis-

tent with a recent publication that reported that, at certain concentrations, OTS167 was capable of

inhibiting the mitotic kinases Aurora B, Haspin, and Bub1 (Ji et al., 2016). We conclude that the

anti-proliferative effects of OTS167 are not a result of its inhibition of MELK.

Generation and characterization of MELK-knockout clonal cell linesTo unambiguously demonstrate that MELK is dispensable for the proliferation of certain cancer cell

lines, we used CRISPR to generate clonal cell lines that lack MELK protein. We transduced the MDA-

MB-231 triple-negative breast cancer cell line with sets of 2 guide RNAs and then expanded clonal

populations from single cells (Figure 4A). Recombination between the chosen gRNA cut sites elimi-

nates exon 3, which encodes residues that are essential for ATP binding (Cao et al., 2013;

Cho et al., 2014), as well as parts of exons 2, 4, and/or 5. We used PCR to identify three indepen-

dent clones that were homozygous for CRISPR-induced recombination, and then confirmed the loss

of the intervening genetic material by sequencing across the gRNA cut sites (Figure 4B–C). Addi-

tionally, we derived clones of Cal51 that have been transduced with single MELK gRNAs, and then

identified three clones that harbored indels in the MELK kinase domain (Figure 4—figure supple-

ment 1). Western blot analysis of the MDA-MB-231 and Cal51 clones with two antibodies that rec-

ognize distinct regions of MELK further verified the complete lack of MELK expression in all six

derived cell lines (Figure 4D–E and Figure 4—figure supplement 1).

The MDA-MB-231and Cal51 MELK-KO clones exhibited robust proliferation, demonstrating that

MELK is fully dispensable for the growth of these cancer cell lines (Figure 4F and Figure 4—figure

supplement 1). In fact, one MDA-MB-231 MELK-KO clone exhibited a significantly shorter doubling

time than the Rosa26 gRNA-transduced cell lines, potentially due to the presence of additional

mutations that were acquired during clonal expansion. The MELK-KO clones progressed through the

cell cycle without gross abnormalities and accumulated few multinucleate cells (Figure 4G and Fig-

ure 4—figure supplement 1). OTS167 treatment of MELK-KO clones caused the formation of multi-

nucleate cells, demonstrating that this drug blocks cytokinesis by inhibiting another cellular target

(Figure 4H and Figure 4—figure supplement 1). Finally, serial dilution analysis revealed that the

MDA-MB-231 and Cal51 MELK-KO clones exhibited equivalent OTS167 GI50 values compared to

Rosa26 gRNA-transduced lines (Figure 4I–J and Figure 4—figure supplement 1). We conclude that

MELK is not an absolute requirement for triple-negative breast cancer proliferation, and that

OTS167 blocks growth in a MELK-independent manner.

DiscussionAs a mitotic kinase highly expressed in many cancer types, MELK has been identified as a promising

target for therapeutic intervention. However, through the use of CRISPR/Cas9-mediated mutagene-

sis, we have demonstrated that MELK is dispensable for growth in 13 out of 13 cancer cell lines

tested, and that a MELK inhibitor currently in clinical trials blocks cell division by inhibiting another

target. We believe that our results highlight the importance of using CRISPR/Cas9 technology to

study and validate preclinical targets in cancer drug development.

Previous research utilizing RNA interference to knock down MELK has indicated that MELK

expression is required for cancer cell proliferation. However, a growing body of evidence has

revealed that RNAi is prone to pervasive off-target effects. This problem is particularly challenging

Figure 3 continued

(E) Summary of GI50 values from OTS167 treatment of MDA-MB-231 cells harboring guide RNAs targeting Rosa26 or MELK. (F) 7 point dose-response

curves of OTS167 in the indicated MDA-MB-231 cell lines.

DOI: 10.7554/eLife.24179.011

The following figure supplements are available for figure 3:

Figure supplement 1. Receptor-positive breast cancer cell lines are sensitive to OTS167.

DOI: 10.7554/eLife.24179.012

Figure supplement 2. OTS167 treatment, but not MELK mutation, causes the accumulation of multinucleate cells.

DOI: 10.7554/eLife.24179.013

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 9 of 17

Short report Cancer Biology Genes and Chromosomes

Rosa

26 g

1

Rosa

26 g

1 +

1µg/m

l Cyt

o B

Rosa

26 g

1 +

2µg/m

l Cyt

o B

Rosa

26 g

1 +

10nM

OTS

167

Rosa

26 g

1 +

30nM

OTS

167

Rosa

26 g

1 +

100n

M O

TS16

7

Rosa

26 g

2

Rosa

26 g

3

MELK

-KO g

1/g5

MELK

-KO g

1/g5

+ 10

0nM

OTS

167

MELK

-KO g

3/g5

MELK

-KO g

3/g5

+ 10

0nM

OTS

167

MELK

-KO g

1/g6

MELK

-KO g

1/g6

+ 10

0nM

OTS

167

0.0

0.2

0.4

0.6

0.8

1.0

Fra

cti

on

of

cells

MDA-MB-231 MELK-KO

Mononucleate

Multinucleate

Rosa

26 g

RNA

MELK

-KO

0

5

10

15

20

GI5

0 v

alu

es (

nM

)

MDA-MB-231: OTSSP167 sensitivity

Cut Site

Deletion

Ro

sa

26

g1

ME

LK

-KO

g1

/g6

Ro

sa

26

g1

ME

LK

-KO

g3

/g5

ME

LK

-KO

g1

/g5

MELKMELK

α-MELK (N-terminal) α-MELK (C-terminal)

alpha-TUBR

osa

26

g1

Ro

sa

26

g2

ME

LK

-KO

g1

/g5

ME

LK

-KO

g3

/g5

ME

LK

-KO

g1

/g6

alpha-TUB

Ro

sa

26

g1

Ro

sa

26

g2

ME

LK

-KO

g1

/g5

ME

LK

-KO

g3

/g5

ME

LK

-KO

g1

/g6

Exon 2 Exon 4

ME

LK

-KO

g1

/g5

Exon 5 Exon 2

ME

LK

-KO

g1

/g6

Exon 2 Complex

ME

LK

-KO

g3

/g5

A B C

D

Rosa

26 g

RNA

MELK

-KO

0

10

20

30

40

Do

ub

lin

g t

ime (

ho

urs

)

MDA-MB-231: Doubling Time

0 1 2 3 4 50

5

10

15

Passage

Po

pu

lati

on

Do

ub

lin

gs

MDA-MB-231: Proliferation

Rosa26 g1

Rosa26 g2

Rosa26 g3

MELK-KO g3/g5

MELK-KO g1/g6

MELK-KO g1/g5

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 Rosa26 g1

GI50 = 11nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 Rosa26 g2

GI50 = 12nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 Rosa26 g3

GI50 = 11nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 MELK-KO g1/g6

GI50 = 10nM

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 MELK-KO g3/g5

GI50 = 11nM

E

Rosa26 g1

Rosa26 g1

+ 100nM OTS167

MELK-KO g3/g5

+ 100nM OTS167MELK-KO g3/g5F G

I HJ

0.1 1 10 100 10000.0

0.5

1.0

[OTSSP167] (nM)

Fra

cti

on

al su

rviv

al

MDA-MB-231 MELK-KO g1/g5

GI50 = 13nM

Figure 4. MELK-knockout cell lines proliferate at normal rates and remain sensitive to OTS167. (A) Schematic of exons in the MDA-MB-231 MELK-KO

g1/g6 knockout line. Half-arrows indicate positions of either cut-site or deletion-spanning primers used to screen these colonies. Primer sequences are

presented in Supplementary file 2. (B) PCR validation of 3 independent MELK-KO clones. Note that amplification of the MELK-KO g3/g5 DNA with

deletion-spanning primers yielded deletion products of at least two distinct sizes. (C) Sanger sequence validation of 3 independent MELK-KO clones.

Figure 4 continued on next page

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 10 of 17

Short report Cancer Biology Genes and Chromosomes

when RNAi is used to study putative cell cycle regulators, as multiple publications have reported

that the cell cycle genes RAD51 and MAD2 are unusually sensitive to off-target RNAi inhibition

(Adamson et al., 2012; Hubner et al., 2010; Sigoillot et al., 2012). For instance, in a screen for

genes whose depletion caused a bypass of the spindle assembly checkpoint, 34 of the top 34 candi-

date siRNA’s exhibited off-target down-regulation of Mad2 levels (Sigoillot et al., 2012). Moreover,

the expression of MELK is strongly cell-cycle regulated: MELK levels are typically low in G0/G1, and

peak in mitosis [(Badouel et al., 2010) and our unpublished data]. A genetic or chemical treatment

that induces a G1 arrest would therefore be predicted to down-regulate MELK, potentially con-

founding the analysis of knockdown efficiency. While Cas9 mutagenesis is also susceptible to off-tar-

get editing, to the best of our knowledge, the off-target loci affected by CRISPR are unlikely to

substantially overlap with those that are affected by RNAi. Moreover, sequencing the locus targeted

by Cas9 can provide an unbiased readout of mutagenesis efficiency that is not sensitive to cell state-

dependent expression variability. Finally, unlike RNAi, CRISPR can be applied to generate clonal cell

lines that harbor null mutations in a targeted gene. This technique bypasses the problems inherent

in the analysis of mixed cell populations and partial loss-of-function phenotypes, and can provide

significant insight into the genetic architecture of cancer.

One limitation of CRISPR mutagenesis is that, over the time required to generate or select for a

pure cell population, cells may engage compensatory mechanisms to buffer against the loss of a tar-

geted protein. Thus, the analysis of knockout clones can be complemented with cell-cell competition

assays, which allow less time for cells to adapt to gene loss and may reveal the presence of a tran-

sient or immediate fitness defect induced by CRISPR. We performed a total of 91 competition assays

(7 MELK gRNAs in 13 different cell lines) that failed to reveal an effect of MELK loss on cell fitness,

further strengthening our conclusion that MELK is dispensable for cancer cell proliferation.

CRISPR mutagenesis can also assist in the pharmacological study of potential drugs. Several lines

of evidence indicate that OTS167 does indeed inhibit MELK: for instance, a crystal structure of

OTS167 binding to the MELK kinase domain has been reported (Cho et al., 2014). However, these

structural and biochemical studies are unable to conclusively demonstrate that a phenotype in a liv-

ing cell is due to an on-target effect. We believe that CRISPR represents a useful tool to gain genetic

insight into this question: if a CRISPR-induced null mutation of a putative drug target fails to confer

resistance to that drug, then that drug must act through alternate targets or mechanisms. While the

MELK-KO cell lines that we generated remain exquisitely sensitive to OTS167, at present, we do not

know how OTS167 blocks cell division. One possibility, not ruled out by our studies, is that OTS167

exhibits polypharmacology (Knight et al., 2010), and kills cancer cells by inhibiting multiple kinases,

potentially including MELK. The analysis of drug-resistant alleles of other mitotic kinases that

OTS167 has been shown to inhibit (Ji et al., 2016) may shed further light on the in vivo MOA of this

compound.

Our results leave open the question of what role, if any, MELK plays in mammalian biology and

cell cycle progression. While MELK is up-regulated in diverse tumor types, it is also expressed in sev-

eral normal cell lineages, including embryonic cells, hematopoietic cells, and neural progenitor cells

(Heyer et al., 1997; Nakano et al., 2005; Gil et al., 1997). MELK may be required at a certain

developmental stage, or for a specific cell type or organismal process. Similarly, we cannot currently

rule out the possibility that MELK plays a role in tumorigenesis in vivo that was not assessed in our

Figure 4 continued

While MELK-KO g1/g6 and g1/g5 harbor a single homozygous deletion, MELK-KO g3/g5 harbors at least two distinct deletions. (D) Western blot

analysis of MELK-KO clones using an antibody that recognizes a region in the N-terminal kinase domain (Abcam ab108529). (E) Western blot analysis of

MELK-KO clones using an antibody that recognizes a region in the C-terminal domain (Cell Signal 2274S). (F) Proliferation analysis and doubling time

measurements of MELK-KO cell lines. (G) Representative images of Rosa26 gRNA or MELK-KO clones either untreated or treated with 100 nM OTS167

and then stained with Hoechst dye. (H) The indicated cell lines were either left untreated or were treated with the cytokinesis inhibitor cytochalasin B or

with OTS167. Cells were then stained with Hoechst dye. For each experiment, at least 200 cells were counted. (I) Summary of GI50 values from OTS167

treatment of either MDA-MB-231 Rosa26 gRNA or MELK-KO clones. (J) 7 point dose-response curves of OTS167 in the indicated cell lines.

DOI: 10.7554/eLife.24179.014

The following figure supplement is available for figure 4:

Figure supplement 1. Generation and analysis of Cal51 MELK-KO cell lines.

DOI: 10.7554/eLife.24179.015

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 11 of 17

Short report Cancer Biology Genes and Chromosomes

current work. At a minimum, our results suggest that MELK is dispensable for mitotic progression in

most cancers. MELK may function in an overlapping or redundant pathway with other mitotic kin-

ases, several of which are up-regulated along with MELK in tumor cells (Malumbres and Barbacid,

2007). Synthetic lethal screens and further in vivo investigation will shed light on MELK’s function in

development and cancer. Nonetheless, our data suggest that specific MELK inhibitors are unlikely to

be useful monoagents in cancer therapy.

Materials and methods

Tissue cultureThe identity of every human cell line utilized in this paper was authenticated using STR profiling (Uni-

versity of Arizona Genetics Core, Tucson, AZ). Cell lines were also confirmed to be negative for

mycoplasma contamination using the MycoAlert Detection Kit (Lonza, Switzerland; LT07-218). Cell

lines HCC70, HCC1937, MDA-MB-453, MDA-MB-468, and ZR-75–1 were grown in RPMI 1640 sup-

plemented with 10% fetal bovine (FBS), 2 mM glutamine, 1% Nonessential Amino Acids (Life

Technologies, Waltham, MA; BY00148), and 100 U/ml penicillin and streptomycin. HCC1143 and

NCI-H1299 were grown in RPMI 1640 supplemented with 10% FBS, 2 mM glutamine, and 100 U/ml

penicillin and streptomycin. T24 was grown in McCoy’s 5A media supplemented with 10% FBS, 2

mM glutamine, and 100 U/ml penicillin and streptomycin. Cal51, A375, MDA-MB-231, HCT116,

Cama1, JIMT-1, and U118-MG cells were grown in DMEM supplemented with 10% FBS, 2 mM gluta-

mine, and 100 U/ml penicillin and streptomycin. T47D cells were grown in RPMI supplemented with

10% FBS, 6.94 mg/ml insulin (Thermo Fisher, Waltham, MA; BN00226), 2 mM glutamine, and 100 U/

ml penicillin and streptomycin. MCF7 cells were grown in DMEM supplemented with 10% FBS, 0.01

mg/ml insulin, 2 mM glutamine, and 100 U/ml penicillin and streptomycin. Cell lines were kindly pro-

vided by the individuals thanked in the acknowledgments. All cell lines were maintained in a humidi-

fied environment at 37˚C and 5% CO2. Cell counting was performed using the Cellometer Auto T4

system (Nexcelom, Lawrence, MA).

Plasmid constructionGuide RNAs targeting protein domains in MELK, PCNA, and RPA3 were designed with assistance

from Osama El Demerdash (manuscript in preparation). Oligonucleotides were ordered from IDT

and then cloned into the LRG 2.1 vector [a gift from Jun-Wei Shi (University of Pennsylvania) and

Chris Vakoc (Cold Spring Harbor Laboratory)] using a BsmBI digestion (Shalem et al., 2014). Plas-

mids were amplified in Stbl3 E. coli (Thermo Fisher; C737303) prepared using the Mix and Go trans-

formation kit (Zymo Research, Irvine, CA; T3001).

Plasmid transfection and transductionHEK293T cells were transfected using the calcium-phosphate method (Smale, 2010). Supernatant

was harvested 48 to 72 hr post-transfection, filtered through a 0.45 mm syringe, and then frozen at

�80˚ C for later use or applied directly to cells with 4 mg/mL polybrene. The culture media on target

plates was changed 24 hr post-transduction.

Proliferation analysisTo measure cell proliferation, 100,000 cells of each strain were plated on a six well plate and then

allowed to grow for 72 hr. Cells were then trypsinized, counted, and 100,000 cells were re-plated in

fresh media. Cells were passaged five times, and cumulative population doublings and doubling

times were calculated at each passage.

Soft agar assaysSolutions of 1.0% and 0.7% Difco Agar Noble in sterile water were autoclaved and then allowed to

equilibrate at 42˚C or 37.5˚C, respectively. The 1% agar solution was then mixed 1:1 with 2 ml of the

appropriate media, supplemented with 2X the concentration of serum, glutamine, and penicillin/

streptomycin. 1 mL of this mixture was then plated in one well of a six well plate and allowed to

solidify at room temperature for 30 min, resulting in a 0.5% agar base layer. Cells of interest were

then trypsinized and re-suspended in their appropriate media. 20,000 cells were diluted in 1 ml of

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 12 of 17

Short report Cancer Biology Genes and Chromosomes

2X media and then mixed with 1 ml of the 0.7% agar solution. 500 ml of this mixture was then plated

on the solidified base layer, resulting in 5000 cells in a 0.375% agar suspension. The agar was

allowed to solidify at room temperature for one hour before being transferred to an incubator. Fresh

1X media was added to the surface of each well 24 hr after plating, and then were re-fed every 3

days. After 21 days of growth, colonies were scored under 10x magnification. All experiments were

plated in triplicate and performed twice.

Western blot analysisCells were lysed with RIPA buffer [25 mM Tris, pH 7.4, 150 mM NaCl, 1% Triton X 100, 0.5% sodium

deoxycholate, 0.1% sodium dodecyl sulfate, protease inhibitor cocktail (Roche, Indianapolis, Indi-

ana), phosphatase inhibitor cocktail (Roche)]. Lysates were quantified using the Pierce BCA Kit

(Thermo Scientific), and equal amounts of protein were denatured and loaded onto an 8% SDS-

PAGE gel. The protein was transferred onto a polyvinylidene difluoride membrane using the Trans-

Blot Turbo Transfer System (Bio-Rad, Hercules, California). The membrane was blocked in 5% non-

fat milk-TBST and then incubated with anti-MELK (abcam, Cambridge, MA; ab108529) at a 1:3000

dilution, or blocked in 10% BSA-TBST and then incubated with anti-MELK (Cell Signal, Danvers, MA;

2274S) at a 1:1000 dilution. Anti-alpha-tubulin (Sigma-Aldrich, St. Louis, MO; T6199) was used as a

loading control at a 1:3000 dilution. All primary antibody incubations were performed overnight at

4˚C. Following incubation, the membranes were washed and then incubated in either anti-rabbit sec-

ondary (abcam; ab6721) at 1:50000 for MELK or anti-mouse secondary (Bio-Rad; 1706516) at

1:10000 for Tubulin for 1 hr at room temperature.

Analysis of CRISPR-mediated mutagenesisGenomic DNA was extracted from transduced cell lines using the QIAmp DNA Mini kit

(Qiagen, Germantown, MD; Cat. No. 51304). Loci targeted by guide RNAs were amplified using the

primers listed in Supplementary file 2, and then sequenced using the forward and reverse primers

at the Cold Spring Harbor Laboratory sequencing facility. Sequence traces were analyzed using TIDE

(Brinkman et al., 2014).

Analysis of OTSSP167 sensitivityFor every cell line of interest, 10,000 cells were plated in 100 ml of media in an 8 � 4 matrix on a flat-

bottomed 96-well plate. Cells were allowed to attach for 24 hr, at which point the media in every

well was changed. 500 nM of OTSSP167 (MedChem Express, Monmouth Junction, NJ; Cat. No. HY-

15512A) was added to one row of cells, and then 6 3-fold serial dilutions were performed. After 72

hr of growth in the presence of the drug, cells were trypsinized and counted using a MacsQuant

Analyzer 10 (Milltenyi Biotec, Germany). The fraction of cells recovered at every drug concentration,

relative to a row of untreated cells, was determined. GI50 values were calculated using a four-param-

eter inhibition vs. concentration model in Prism 7 (Graphpad, San Diego, California). Sensitivity

experiments in Figure 4J and Figure 3—figure supplement 1 were performed 2–3 times each,

while sensitivity experiments in Figure 3 and Figure 4—figure supplement 1 were performed once.

GFP dropout screeningCells were transduced on day 0 with sgRNA lentiviral supernatant, which was then replaced with

fresh media on day 1. On day 3, the baseline percentage of GFP+ cells was measured using a Macs-

Quant Analyzer 10 (Milltenyi Biotec). Cells were then passaged every 3 or 4 days, according to their

growth rate and confluence, and the percentage of GFP+ cells was measured at every split. Dropout

values represent the fold decrease in GFP+ cells at each passage, relative to the GFP+ percentage

on day 3. In preliminary experiments with A375 and MDA-MB-231, replicate dropout assays were

highly reproducible across independent replicates. For that reason, GFP dropout experiments in the

13 tested cell lines were performed once.

DNA staining10,000 (A375, Cal51) or 20,000 (MDA-MB-231) cells of interest were plated in 250 ml of media in a

flat-bottomed 24-well plate and allowed to attach for 24 hr. Then the media was replaced, and

OTS167 or Cytochalasin B (Cayman Chemical Company, Ann Arbor, MI; Cat. No. 11328) were added

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 13 of 17

Short report Cancer Biology Genes and Chromosomes

to control wells. Following an additional 24 hr period of growth, cells were stained with 2.5 mg/ml of

Hoechst dye (Thermo Fisher, Cat. No. H3569) for 30 min and imaged using appropriate filters. DNA

staining experiments were performed twice.

AcknowledgementsWe thank Chris Vakoc and Junwei Shi for providing Cas9 and guide RNA expression plasmids. We

thank Chris Vakoc, Camila dos Santos, and Ken Chang for providing cell lines. We thank Allison Leo-

nard for technical assistance with lentiviral production. We thank Osama El Demerdash for assistance

designing domain-targeted guide RNAs. This work was performed with assistance from CSHL

Shared Resources, including the CSHL Flow Cytometry Shared Resource, which are supported by

the Cancer Center Support Grant 5P30CA045508. Research in the Sheltzer Lab is supported by an

NIH Early Independence Award (1DP5OD021385).

Additional information

Funding

Funder Grant reference number Author

National Institutes of Health 1DP5OD021385 Jason M Sheltzer

The funders had no role in study design, data collection and interpretation, or the decision tosubmit the work for publication.

Author contributions

AL, CJG, NMS, Investigation, Methodology; JMS, Conceptualization, Supervision, Funding acquisi-

tion, Investigation, Methodology, Writing—original draft, Writing—review and editing

Author ORCIDs

Ann Lin, http://orcid.org/0000-0002-4618-8120

Christopher J Giuliano, http://orcid.org/0000-0002-0586-6095

Nicole M Sayles, http://orcid.org/0000-0001-7460-9095

Jason M Sheltzer, http://orcid.org/0000-0003-1381-1323

Additional filesSupplementary files. Supplementary file 1. Guide RNA sequences. The sequences of every guide RNA and the protein

domain targeted by the guide RNAs are displayed.

DOI: 10.7554/eLife.24179.016

. Supplementary file 2. PCR primers to amplify MELK gRNA cut sites and deletions. The sequences

of PCR primers used in this manuscript are displayed.

DOI: 10.7554/eLife.24179.017

ReferencesAdamson B, Smogorzewska A, Sigoillot FD, King RW, Elledge SJ. 2012. A genome-wide homologousrecombination screen identifies the RNA-binding protein RBMX as a component of the DNA-damage response.Nature Cell Biology 14:318–328. doi: 10.1038/ncb2426, PMID: 22344029

Aguirre AJ, Meyers RM, Weir BA, Vazquez F, Zhang CZ, Ben-David U, Cook A, Ha G, Harrington WF, Doshi MB,Kost-Alimova M, Gill S, Xu H, Ali LD, Jiang G, Pantel S, Lee Y, Goodale A, Cherniack AD, Oh C, et al. 2016.Genomic copy number dictates a Gene-Independent cell response to CRISPR/Cas9 targeting. Cancer Discovery6:914–929. doi: 10.1158/2159-8290.CD-16-0154, PMID: 27260156

Alachkar H, Mutonga MB, Metzeler KH, Fulton N, Malnassy G, Herold T, Spiekermann K, Bohlander SK,Hiddemann W, Matsuo Y, Stock W, Nakamura Y. 2014. Preclinical efficacy of maternal embryonic leucine-zipper

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 14 of 17

Short report Cancer Biology Genes and Chromosomes

kinase (MELK) inhibition in acute myeloid leukemia. Oncotarget 5:12371–12382. doi: 10.18632/oncotarget.2642, PMID: 25365263

Badouel C, Chartrain I, Blot J, Tassan JP. 2010. Maternal embryonic leucine zipper kinase is stabilized in mitosisby phosphorylation and is partially degraded upon mitotic exit. Experimental Cell Research 316:2166–2173.doi: 10.1016/j.yexcr.2010.04.019, PMID: 20420823

Badve S, Dabbs DJ, Schnitt SJ, Baehner FL, Decker T, Eusebi V, Fox SB, Ichihara S, Jacquemier J, Lakhani SR,Palacios J, Rakha EA, Richardson AL, Schmitt FC, Tan PH, Tse GM, Weigelt B, Ellis IO, Reis-Filho JS. 2011.Basal-like and triple-negative breast cancers: a critical review with an emphasis on the implications forpathologists and oncologists. Modern Pathology 24:157–167. doi: 10.1038/modpathol.2010.200,PMID: 21076464

Beke L, Kig C, Linders JT, Boens S, Boeckx A, van Heerde E, Parade M, De Bondt A, Van den Wyngaert I, BashirT, Ogata S, Meerpoel L, Van Eynde A, Johnson CN, Beullens M, Brehmer D, Bollen M. 2015. MELK-T1, a small-molecule inhibitor of protein kinase MELK, decreases DNA-damage tolerance in proliferating Cancer cells.Bioscience Reports 35:e00267. doi: 10.1042/BSR20150194, PMID: 26431963

Boettcher M, McManus MT. 2015. Choosing the right tool for the job: rnai, TALEN, or CRISPR. Molecular Cell58:575–585. doi: 10.1016/j.molcel.2015.04.028, PMID: 26000843

Brinkman EK, Chen T, Amendola M, van Steensel B. 2014. Easy quantitative assessment of genome editing bysequence trace decomposition. Nucleic Acids Research 42:e168. doi: 10.1093/nar/gku936, PMID: 25300484

Campbell J, Ryan CJ, Brough R, Bajrami I, Pemberton HN, Chong IY, Costa-Cabral S, Frankum J, Gulati A,Holme H, Miller R, Postel-Vinay S, Rafiq R, Wei W, Williamson CT, Quigley DA, Tym J, Al-Lazikani B, Fenton T,Natrajan R, et al. 2016. Large-Scale profiling of kinase dependencies in Cancer cell lines. Cell Reports 14:2490–2501. doi: 10.1016/j.celrep.2016.02.023, PMID: 26947069

Cao LS, Wang J, Chen Y, Deng H, Wang ZX, Wu JW. 2013. Structural basis for the regulation of maternalembryonic leucine zipper kinase. PLoS One 8:e70031. doi: 10.1371/journal.pone.0070031, PMID: 23922895

Cho YS, Kang Y, Kim K, Cha YJ, Cho HS. 2014. The crystal structure of MPK38 in complex with OTSSP167, anorally administrative MELK selective inhibitor. Biochemical and Biophysical Research Communications 447:7–11. doi: 10.1016/j.bbrc.2014.03.034, PMID: 24657156

Choi S, Ku JL, J-l K. 2011. Resistance of colorectal Cancer cells to radiation and 5-FU is associated with MELKexpression. Biochemical and Biophysical Research Communications 412:207–213. doi: 10.1016/j.bbrc.2011.07.060, PMID: 21806965

Chung S, Suzuki H, Miyamoto T, Takamatsu N, Tatsuguchi A, Ueda K, Kijima K, Nakamura Y, Matsuo Y. 2012.Development of an orally-administrative MELK-targeting inhibitor that suppresses the growth of various typesof human Cancer. Oncotarget 3:1629–1640. doi: 10.18632/oncotarget.790, PMID: 23283305

Cowley GS, Weir BA, Vazquez F, Tamayo P, Scott JA, Rusin S, East-Seletsky A, Ali LD, Gerath WF, Pantel SE,Lizotte PH, Jiang G, Hsiao J, Tsherniak A, Dwinell E, Aoyama S, Okamoto M, Harrington W, Gelfand E, GreenTM, et al. 2014. Parallel genome-scale loss of function screens in 216 Cancer cell lines for the identification ofcontext-specific genetic dependencies. Scientific Data 1:140035. doi: 10.1038/sdata.2014.35, PMID: 25984343

Du T, Qu Y, Li J, Li H, Su L, Zhou Q, Yan M, Li C, Zhu Z, Liu B. 2014. Maternal embryonic leucine zipper kinaseenhances gastric Cancer progression via the FAK/Paxillin pathway. Molecular Cancer 13:100. doi: 10.1186/1476-4598-13-100, PMID: 24885567

Gil M, Yang Y, Lee Y, Choi I, Ha H. 1997. Cloning and expression of a cDNA encoding a novel protein serine/threonine kinase predominantly expressed in hematopoietic cells. Gene 195:295–301. doi: 10.1016/S0378-1119(97)00181-9, PMID: 9305775

Gray D, Jubb AM, Hogue D, Dowd P, Kljavin N, Yi S, Bai W, Frantz G, Zhang Z, Koeppen H, de Sauvage FJ,Davis DP. 2005. Maternal embryonic leucine zipper kinase/murine protein serine-threonine kinase 38 is apromising therapeutic target for multiple cancers. Cancer Research 65:9751–9761. doi: 10.1158/0008-5472.CAN-04-4531, PMID: 16266996

Gu C, Banasavadi-Siddegowda YK, Joshi K, Nakamura Y, Kurt H, Gupta S, Nakano I. 2013. Tumor-Specificactivation of the C-JUN/MELK pathway regulates glioma stem cell growth in a p53-Dependent manner. StemCells 31:870–881. doi: 10.1002/stem.1322

Hart T, Brown KR, Sircoulomb F, Rottapel R, Moffat J. 2014. Measuring error rates in genomic perturbationscreens: gold standards for human functional genomics. Molecular Systems Biology 10:733. doi: 10.15252/msb.20145216, PMID: 24987113

Hart T, Chandrashekhar M, Aregger M, Steinhart Z, Brown KR, MacLeod G, Mis M, Zimmermann M, Fradet-Turcotte A, Sun S, Mero P, Dirks P, Sidhu S, Roth FP, Rissland OS, Durocher D, Angers S, Moffat J. 2015. High-Resolution CRISPR screens reveal fitness genes and Genotype-Specific Cancer liabilities. Cell 163:1515–1526.doi: 10.1016/j.cell.2015.11.015, PMID: 26627737

Hebbard LW, Maurer J, Miller A, Lesperance J, Hassell J, Oshima RG, Terskikh AV. 2010. Maternal embryonicleucine zipper kinase is upregulated and required in mammary tumor-initiating cells in vivo. Cancer Research70:8863–8873. doi: 10.1158/0008-5472.CAN-10-1295, PMID: 20861186

Heyer BS, Warsowe J, Solter D, Knowles BB, Ackerman SL. 1997. New member of the Snf1/AMPK kinase family,melk, is expressed in the mouse egg and preimplantation embryo. Molecular Reproduction and Development47:148–156. doi: 10.1002/(SICI)1098-2795(199706)47:2<148::AID-MRD4>3.0.CO;2-M, PMID: 9136115

Hubner NC, Wang LH, Kaulich M, Descombes P, Poser I, Nigg EA. 2010. Re-examination of siRNA specificityquestions role of PICH and Tao1 in the spindle checkpoint and identifies Mad2 as a sensitive target for smallRNAs. Chromosoma 119:149–165. doi: 10.1007/s00412-009-0244-2, PMID: 19904549

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 15 of 17

Short report Cancer Biology Genes and Chromosomes

Jackson AL, Bartz SR, Schelter J, Kobayashi SV, Burchard J, Mao M, Li B, Cavet G, Linsley PS. 2003. Expressionprofiling reveals off-target gene regulation by RNAi. Nature Biotechnology 21:635–637. doi: 10.1038/nbt831,PMID: 12754523

Jackson AL, Burchard J, Schelter J, Chau BN, Cleary M, Lim L, Linsley PS. 2006. Widespread siRNA "off-target"transcript silencing mediated by seed region sequence complementarity. RNA 12:1179–1187. doi: 10.1261/rna.25706, PMID: 16682560

Ji W, Arnst C, Tipton AR, Bekier ME, Taylor WR, Yen TJ, Liu ST. 2016. OTSSP167 abrogates mitotic checkpointthrough inhibiting multiple mitotic kinases. PLoS One 11:e0153518. doi: 10.1371/journal.pone.0153518,PMID: 27082996

Johnson CN, Berdini V, Beke L, Bonnet P, Brehmer D, Coyle JE, Day PJ, Frederickson M, Freyne EJ, Gilissen RA,Hamlett CC, Howard S, Meerpoel L, McMenamin R, Patel S, Rees DC, Sharff A, Sommen F, Wu T, Linders JT.2015a. Fragment-based discovery of type I inhibitors of maternal embryonic leucine zipper kinase. ACSMedicinal Chemistry Letters 6:25–30. doi: 10.1021/ml5001245, PMID: 25589925

Johnson CN, Adelinet C, Berdini V, Beke L, Bonnet P, Brehmer D, Calo F, Coyle JE, Day PJ, Frederickson M,Freyne EJ, Gilissen RA, Hamlett CC, Howard S, Meerpoel L, Mevellec L, McMenamin R, Pasquier E, Patel S,Rees DC, et al. 2015b. Structure-Based design of type II inhibitors applied to maternal embryonic leucinezipper kinase. ACS Medicinal Chemistry Letters 6:31–36. doi: 10.1021/ml5001273, PMID: 25589926

Jung H, Seong HA, Ha H. 2008. Murine protein serine/threonine kinase 38 activates apoptosis signal-regulatingkinase 1 via thr 838 phosphorylation. Journal of Biological Chemistry 283:34541–34553. doi: 10.1074/jbc.M807219200, PMID: 18948261

Kig C, Beullens M, Beke L, Van Eynde A, Linders JT, Brehmer D, Bollen M. 2013. Maternal embryonic leucinezipper kinase (MELK) reduces replication stress in glioblastoma cells. Journal of Biological Chemistry 288:24200–24212. doi: 10.1074/jbc.M113.471433, PMID: 23836907

Knight ZA, Lin H, Shokat KM. 2010. Targeting the Cancer kinome through polypharmacology. Nature ReviewsCancer 10:130–137. doi: 10.1038/nrc2787, PMID: 20094047

Kuner R, Falth M, Pressinotti NC, Brase JC, Puig SB, Metzger J, Gade S, Schafer G, Bartsch G, Steiner E, KlockerH, Sultmann H. 2013. The maternal embryonic leucine zipper kinase (MELK) is upregulated in high-gradeprostate Cancer. Journal of Molecular Medicine 91:237–248. doi: 10.1007/s00109-012-0949-1, PMID: 22945237

Lin ML, Park JH, Nishidate T, Nakamura Y, Katagiri T. 2007. Involvement of maternal embryonic leucine zipperkinase (MELK) in mammary carcinogenesis through interaction with Bcl-G, a pro-apoptotic member of the Bcl-2family. Breast Cancer Research 9:R17. doi: 10.1186/bcr1650, PMID: 17280616

Luo J, Solimini NL, Elledge SJ. 2009. Principles of Cancer therapy: oncogene and non-oncogene addiction. Cell136:823–837. doi: 10.1016/j.cell.2009.02.024, PMID: 19269363

Malumbres M, Barbacid M. 2007. Cell cycle kinases in Cancer. Current Opinion in Genetics & Development 17:60–65. doi: 10.1016/j.gde.2006.12.008, PMID: 17208431

Marcotte R, Brown KR, Suarez F, Sayad A, Karamboulas K, Krzyzanowski PM, Sircoulomb F, Medrano M,Fedyshyn Y, Koh JL, van Dyk D, Fedyshyn B, Luhova M, Brito GC, Vizeacoumar FJ, Vizeacoumar FS, Datti A,Kasimer D, Buzina A, Mero P, et al. 2012. Essential gene profiles in breast, pancreatic, and ovarian Cancer cells.Cancer Discovery 2:172–189. doi: 10.1158/2159-8290.CD-11-0224, PMID: 22585861

Marcotte R, Sayad A, Brown KR, Sanchez-Garcia F, Reimand J, Haider M, Virtanen C, Bradner JE, Bader GD,Mills GB, Pe’er D, Moffat J, Neel BG. 2016. Functional genomic landscape of human breast Cancer drivers,vulnerabilities, and resistance. Cell 164:293–309. doi: 10.1016/j.cell.2015.11.062, PMID: 26771497

Marie SK, Okamoto OK, Uno M, Hasegawa AP, Oba-Shinjo SM, Cohen T, Camargo AA, Kosoy A, Carlotti CG,Toledo S, Moreira-Filho CA, Zago MA, Simpson AJ, Caballero OL. 2008. Maternal embryonic leucine zipperkinase transcript abundance correlates with malignancy grade in human astrocytomas. International Journal ofCancer 122:807–815. doi: 10.1002/ijc.23189, PMID: 17960622

Mohr SE, Smith JA, Shamu CE, Neumuller RA, Perrimon N. 2014. RNAi screening comes of age: improvedtechniques and complementary approaches. Nature Reviews Molecular Cell Biology 15:591–600. doi: 10.1038/nrm3860, PMID: 25145850

Nakano I, Paucar AA, Bajpai R, Dougherty JD, Zewail A, Kelly TK, Kim KJ, Ou J, Groszer M, Imura T, Freije WA,Nelson SF, Sofroniew MV, Wu H, Liu X, Terskikh AV, Geschwind DH, Kornblum HI. 2005. Maternal embryonicleucine zipper kinase (MELK) regulates multipotent neural progenitor proliferation. The Journal of Cell Biology170:413–427. doi: 10.1083/jcb.200412115, PMID: 16061694

Nakano I, Masterman-Smith M, Saigusa K, Paucar AA, Horvath S, Shoemaker L, Watanabe M, Negro A, Bajpai R,Howes A, Lelievre V, Waschek JA, Lazareff JA, Freije WA, Liau LM, Gilbertson RJ, Cloughesy TF, GeschwindDH, Nelson SF, Mischel PS, et al. 2008. Maternal embryonic leucine zipper kinase is a key regulator of theproliferation of malignant brain tumors, including brain tumor stem cells. Journal of Neuroscience Research 86:48–60. doi: 10.1002/jnr.21471, PMID: 17722061

Rakha EA, Reis-Filho JS, Ellis IO. 2008. Basal-like breast Cancer: a critical review. Journal of Clinical Oncology 26:2568–2581. doi: 10.1200/JCO.2007.13.1748, PMID: 18487574

Ryu B, Kim DS, Deluca AM, Alani RM. 2007. Comprehensive expression profiling of tumor cell lines identifiesmolecular signatures of melanoma progression. PLoS One 2:e594. doi: 10.1371/journal.pone.0000594,PMID: 17611626

Shalem O, Sanjana NE, Hartenian E, Shi X, Scott DA, Mikkelsen TS, Heckl D, Ebert BL, Root DE, Doench JG,Zhang F. 2014. Genome-scale CRISPR-Cas9 knockout screening in human cells. Science 343:84–87. doi: 10.1126/science.1247005, PMID: 24336571

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 16 of 17

Short report Cancer Biology Genes and Chromosomes

Sheltzer JM. 2013. A transcriptional and metabolic signature of primary aneuploidy is present in chromosomallyunstable Cancer cells and informs clinical prognosis. Cancer Research 73:6401–6412. doi: 10.1158/0008-5472.CAN-13-0749, PMID: 24041940

Shi J, Wang E, Milazzo JP, Wang Z, Kinney JB, Vakoc CR. 2015. Discovery of Cancer drug targets by CRISPR-Cas9 screening of protein domains. Nature Biotechnology 33:661–667. doi: 10.1038/nbt.3235, PMID: 25961408

Sigoillot FD, Lyman S, Huckins JF, Adamson B, Chung E, Quattrochi B, King RW. 2012. A bioinformatics methodidentifies prominent off-targeted transcripts in RNAi screens. Nature Methods 9:363–366. doi: 10.1038/nmeth.1898, PMID: 22343343

Silva JM, Marran K, Parker JS, Silva J, Golding M, Schlabach MR, Elledge SJ, Hannon GJ, Chang K. 2008.Profiling essential genes in human mammary cells by multiplex RNAi screening. Science 319:617–620. doi: 10.1126/science.1149185, PMID: 18239125

Singh S, Narang AS, Mahato RI. 2011. Subcellular fate and off-target effects of siRNA, shRNA, and miRNA.Pharmaceutical Research 28:2996–3015. doi: 10.1007/s11095-011-0608-1, PMID: 22033880

Smale ST. 2010. Calcium phosphate transfection of 3t3 fibroblasts. Cold Spring Harbor Protocols 2010:pdb.prot5372. doi: 10.1101/pdb.prot5372, PMID: 20150135

Speers C, Zhao SG, Kothari V, Santola A, Liu M, Wilder-Romans K, Evans J, Batra N, Bartelink H, Hayes DF,Lawrence TS, Brown PH, Pierce LJ, Feng FY. 2016. Maternal embryonic leucine zipper kinase (MELK) as a novelmediator and biomarker of radioresistance in human breast Cancer. Clinical Cancer Research 22:5864–5875.doi: 10.1158/1078-0432.CCR-15-2711, PMID: 27225691

Sørlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H, Hastie T, Eisen MB, van de Rijn M, Jeffrey SS,Thorsen T, Quist H, Matese JC, Brown PO, Botstein D, Lønning PE, Børresen-Dale AL. 2001. Gene expressionpatterns of breast carcinomas distinguish tumor subclasses with clinical implications. PNAS 98:10869–10874.doi: 10.1073/pnas.191367098, PMID: 11553815

Toure BB, Giraldes J, Smith T, Sprague ER, Wang Y, Mathieu S, Chen Z, Mishina Y, Feng Y, Yan-Neale Y, ShakyaS, Chen D, Meyer M, Puleo D, Brazell JT, Straub C, Sage D, Wright K, Yuan Y, Chen X, et al. 2016. Toward thevalidation of maternal embryonic leucine zipper kinase: discovery, optimization of highly potent and selectiveinhibitors, and preliminary biology insight. Journal of Medicinal Chemistry 59:4711–4723. doi: 10.1021/acs.jmedchem.6b00052, PMID: 27187609

Tzelepis K, Koike-Yusa H, De Braekeleer E, Li Y, Metzakopian E, Dovey OM, Mupo A, Grinkevich V, Li M, MazanM, Gozdecka M, Ohnishi S, Cooper J, Patel M, McKerrell T, Chen B, Domingues AF, Gallipoli P, Teichmann S,Ponstingl H, et al. 2016. A CRISPR dropout screen identifies genetic vulnerabilities and therapeutic targets inacute myeloid leukemia. Cell Reports 17:1193–1205. doi: 10.1016/j.celrep.2016.09.079, PMID: 27760321

Vulsteke V, Beullens M, Boudrez A, Keppens S, Van Eynde A, Rider MH, Stalmans W, Bollen M. 2004. Inhibitionof spliceosome assembly by the cell cycle-regulated protein kinase MELK and involvement of splicing factorNIPP1. Journal of Biological Chemistry 279:8642–8647. doi: 10.1074/jbc.M311466200, PMID: 14699119

Wang Y, Lee YM, Baitsch L, Huang A, Xiang Y, Tong H, Lako A, Von T, Choi C, Lim E, Min J, Li L, Stegmeier F,Schlegel R, Eck MJ, Gray NS, Mitchison TJ, Zhao JJ. 2014. MELK is an oncogenic kinase essential for mitoticprogression in basal-like breast Cancer cells. eLife 3:e01763. doi: 10.7554/eLife.01763, PMID: 24844244

Weinstein IB. 2002. Cancer. addiction to oncogenes–the achilles heal of Cancer. Science 297:63–64. doi: 10.1126/science.1073096, PMID: 12098689

Xia H, Kong SN, Chen J, Shi M, Sekar K, Seshachalam VP, Rajasekaran M, Goh BK, Ooi LL, Hui KM. 2016. MELKis an oncogenic kinase essential for early hepatocellular carcinoma recurrence. Cancer Letters 383:85–93.doi: 10.1016/j.canlet.2016.09.017, PMID: 27693640

Lin et al. eLife 2017;6:e24179. DOI: 10.7554/eLife.24179 17 of 17

Short report Cancer Biology Genes and Chromosomes